COVID-19 detection using lightweight CNN architecture on chest X-ray images

Abstract

Coronavirus disease (Covid-19) detection has been a significant challenge for medical personal's all over the world. Reverse Transcription Polymerase Chain Reaction (RTPCR) is currently utilized to diagnose the Covid-19 disease. However, due to various subjective considerations and ambiguities, the RTPCR test is not a viable option in different circumstances. Radio-graphic images, such as chest X-rays are faster and less expensive than PCR tests while they can provide substantially good results in diagnosing Covid-19. In this research, a Convolutional Neural Network (CNN) model based on depthwise separable convolutions has been proposed to identify Covid-19 from chest X-ray images. Also, various state-of-the-art CNN model has been used and their performance metrics are compared. The analysis indicates that the proposed CNN model can correctly diagnose Covid-19 from the chest X-ray images with a substantially high validation and testing accuracy.